Utah State University
DigitalCommons@USU
All Graduate Theses and Dissertations
Graduate Studies
2016
The Effects of Upstream Straight Pipe Length on
Magnetic Flow Meter Accuracy
Bradley C. Clawson
Utah State University
Follow this and additional works at: http://digitalcommons.usu.edu/etd
Part of the Civil and Environmental Engineering Commons
Recommended Citation
Clawson, Bradley C., "The Effects of Upstream Straight Pipe Length on Magnetic Flow Meter Accuracy" (2016). All Graduate Theses
and Dissertations. Paper 5053.
This Thesis is brought to you for free and open access by the Graduate
Studies at DigitalCommons@USU. It has been accepted for inclusion in All
Graduate Theses and Dissertations by an authorized administrator of
DigitalCommons@USU. For more information, please contact
[email protected].
THE EFFECTS OF UPSTREAM STRAIGHT PIPE LENGTH
ON MAGNETIC FLOW METER ACCURACY
by
Bradley C. Clawson
A thesis submitted in partial fulfillment
of the requirements for the degree
of
MASTER OF SCIENCE
in
Civil and Environmental Engineering
Approved:
____________________________
Steven L. Barfuss
Major Professor
____________________________
Michael C. Johnson
Committee Member
____________________________
John D. Rice
Committee Member
____________________________
Mark R. McLellan
Vice President for Research and
Dean of the School of Graduate Studies
UTAH STATE UNIVERSITY
Logan, Utah
2016
ii
Copyright © Bradley C. Clawson 2016
All Rights Reserved
iii
ABSTRACT
The Effects of Upstream Straight Pipe Length
on Magnetic Flow Meter Accuracy
by
Bradley C. Clawson, Master of Science
Utah State University, 2016
Major Professor: Steven L. Barfuss
Department: Civil & Environmental Engineering
In closed conduit water systems, being able to accurately measure flow is
absolutely essential. For many meter designs, including electromagnetic induction meters
(also known as magnetic flow meters), the greatest accuracy is achieved when the device
is calibrated correctly. Optimal meter accuracy often depends on the flow conditions
associated with the upstream geometry in the pipe system. Manufacturers typically give
standards for the length of straight pipe required upstream of the meter. These standards
vary, however, and do not address every possible configuration that may be installed
upstream of the straight pipe requirement.
An investigation on the effects of the length of straight pipe between a single 90°
elbow and the upstream side of magnetic flow meters was completed in this research.
Eleven 10-inch meters were chosen for testing. The procedure included a baseline test
iv
with more than forty diameters of straight pipe between the elbow and the meter. The
accuracy of the meter was determined over a range of flow velocities typical for
operation of this type and size of meter. Further tests were performed with the meter
installed only three diameters downstream of the elbow. These tests constitute Phase I. In
Phase II, four meters were tested with the upstream pipe length varying from a closecoupled installation to ten diameters of straight pipe between the elbow and the meter to
observe variances in accuracy with distance from the elbow.
The intent of the research was to show whether manufacturer accuracy
specifications are achievable in actual application. It was determined that very few meters
meet the manufacturer’s specification for accuracy even when installation requirements
were met. Post-factory calibrations and minimization of velocity profile disruption
through consideration of upstream geometry is recommended.
(49 Pages)
v
PUBLIC ABSTRACT
The Effects of Upstream Straight Pipe Length
on Magnetic Flow Meter Accuracy
by
Bradley C. Clawson, Master of Science
Utah State University, 2016
Major Professor: Steven L. Barfuss
Department: Civil & Environmental Engineering
In closed conduit water systems, being able to accurately measure flow is
absolutely essential. For many meter designs, including electromagnetic induction meters
(also known as magnetic flow meters), the greatest accuracy is achieved when the device
is calibrated correctly. Optimal meter accuracy often depends on the flow conditions
associated with the upstream geometry in the pipe system. Manufacturers typically give
standards for the length of straight pipe required upstream of the meter. These standards
vary, however, and do not address every possible configuration that may be installed
upstream of the straight pipe requirement.
An investigation on the effects of the length of straight pipe between a single 90°
elbow and the upstream side of magnetic flow meters was done in this research. Eleven
10-inch meters were chosen for testing. The procedure included a baseline test with more
vi
than forty diameters of straight pipe between the elbow and the meter. The accuracy of
the meter was determined over a range of flow velocities typical for operation of this type
and size of meter. Further tests were performed with the meter installed only three
diameters downstream of the elbow. These tests constitute Phase I. In Phase II, four
meters were tested with the upstream pipe length varying from a close-coupled
installation to ten diameters of straight pipe between the elbow and the meter to observe
variances in accuracy with distance from the elbow.
The intent of the research was to show whether manufacturer accuracy
specifications are achievable in actual application. It was determined that very few meters
meet the manufacturer’s specification for accuracy even when installation requirements
were met. Post-factory calibrations and minimization of velocity profile disruption
through consideration of upstream geometry is recommended.
vii
ACKNOWLEDGMENTS
Working as a Research Assistant at the Utah Water Research Laboratory under
Steven Barfuss has opened my field of view, not only to the techniques and subtleties of
Civil Engineering Hydraulics, but also to what I can accomplish as I venture forward
with my career. A special thanks to Steve, as well as Zac Sharp, Dr. Mike Johnson, and
Dr. John Rice for the time and advice contributed to this project. Also, I want to thank my
colleagues, Jordan Jarrett and Jared Justensen for their assistance with testing.
I want to thank my family and friends for encouraging me to continue with my
education. I could not have accomplished this without the immense support of my wife,
who stood by me through my doubts and long hours away from home.
viii
CONTENTS
Page
ABSTRACT....................................................................................................................... iii
PUBLIC ABSTRACT .........................................................................................................v
ACKNOWLEDGMENTS ................................................................................................ vii
CONTENTS..................................................................................................................... viii
LIST OF TABLES ...............................................................................................................x
LIST OF FIGURES ........................................................................................................... xi
NOTATION ..................................................................................................................... xiii
CHAPTER I INTRODUCTION ..........................................................................................1
Purpose ............................................................................................................................ 1
Objective ......................................................................................................................... 2
CHAPTER II LITERATURE REVIEW .............................................................................4
CHAPTER III EXPERIMENTAL SETUP AND PROCEDURE .......................................6
Meters .............................................................................................................................. 6
Experimental Setup ......................................................................................................... 7
Procedure....................................................................................................................... 12
Phase I Test................................................................................................................ 12
Phase II Test .............................................................................................................. 14
CHAPTER IV RESULTS AND DISCUSSION ...............................................................16
Phase I Tests .................................................................................................................. 17
Phase II Tests ................................................................................................................ 27
ix
CHAPTER V CONCLUSION...........................................................................................33
Results Summary........................................................................................................... 33
Need for Further Research ............................................................................................ 34
REFERENCES ..................................................................................................................35
x
LIST OF TABLES
Table
Page
1. Meter Accuracy Specifications and Installation Requirements .......................................7
2. Phase I Test Statistical Data Summary ..........................................................................25
3. Phase II Statistical Data Summary .................................................................................32
xi
LIST OF FIGURES
Figure
Page
1. Example Installation Requirement Schematic .................................................................2
2. Decay of the effect of a single, 90° long radius elbow. (Kelner, 2003)...........................5
3. Phase I Straight Pipe Test Setup (flow left to right) ........................................................8
4. Phase I 3D Pipe Test Setup (flow left to right) ................................................................8
5. Phase II Straight Test Pipe Setup .....................................................................................9
6. Phase II Elbow Test Upstream Pipe Setup ....................................................................10
7. Variations in Pipe Length for Phase II Elbow Test .......................................................11
8. Phase II Straight Pipe Test Setup (flow top to bottom) .................................................11
9. Phase II Elbow Setup for 5D Test (flow left to right)....................................................12
10. Reference Meter Calibration ........................................................................................15
11. Phase I Test Data for Mfr A-1 .....................................................................................18
12. Phase I Test Data for Mfr A-2 .....................................................................................19
13. Phase I Test Data for Mfr B .........................................................................................19
14. Phase I Test Data for Mfr C-1......................................................................................20
15. Phase I Test Data for Mfr C-2......................................................................................20
16. Phase I Test Data for Mfr D .........................................................................................21
17. Phase I Test Data for Mfr E .........................................................................................21
18. Phase I Test Data for Mfr F .........................................................................................22
19. Phase I Test Data for Mfr G .........................................................................................22
20. Phase I Test Data for Mfr H .........................................................................................23
xii
21. Phase I Test Data for Mfr I ..........................................................................................23
22. Phase I 3D Test Deviation From Straight Test ............................................................26
23. Phase II Test Data for Mfr F ........................................................................................28
24. Phase II Test Data for Mfr B........................................................................................28
25. Phase II Test Data for Mfr G .......................................................................................29
26. Phase II Test Data for Mfr H .......................................................................................29
xiii
NOTATION
l
=
length of conductor (meters)
v
=
velocity (meters per second)
B
=
magnetic flux density (Tesla)
e
=
Voltage
Mfr
=
manufacturer
US
=
upstream
DS
=
downstream
1D
=
one diameter
2D
=
two diameters
3D
=
three diameters
5D
=
five diameters
6D
=
six diameters
10D
=
ten diameters
CC
=
close coupled
PVC
=
polyvinyl chloride
cfs
=
cubic feet per second
lbs
=
pounds
s
=
seconds
pcf
=
pounds per cubic foot
Q
=
flow rate (cfs)
W
=
weight of water (lbs)
xiv
t
=
time (s)
γ
=
unit weight of water (pcf)
mA
=
milliamps
Hz
=
hertz
Range =
range of the multimeter
AD
average difference
=
AAD =
adjusted average difference
STDV =
standard deviation
MD
=
max difference
accy
=
accuracy
spec
=
specification
fps
=
feet per second
min
=
minimum
vel
=
velocity
DC
=
direct current
AC
=
alternating current
CHAPTER I
INTRODUCTION
Purpose
In closed conduit water systems, such as those used in irrigation or municipal
water distribution, being able to accurately measure flow is extremely important. Even
small errors when managing large volumes of water will result in either revenue losses or
utility overcharges.
Widespread application of magnetic flowmeters has been available for many
years. In recent years, electromagnetic induction meters (known as magnetic flow meters
or “mag” meters) have improved in accuracy and decreased in cost, according to Thorn
(2000). Magnetic flow meters commonly advertise high accuracy ratings and noninvasive measurement with little or no head loss, making them a competitive option for
closed conduit flow measurement.
For many meter designs, including magnetic flow meters, ideal flow stream is
required to eliminate errors in measurement. “Ideal” flow is defined by Kelner (2003) as
having a fully developed, axisymmetric, swirl free velocity profile. Controlling flow
conditions within the meter depends in part on the upstream geometry of the pipe system.
Manufacturers typically give standards for the length of straight pipe required upstream
of the meter. These standards vary, however, and do not address every case of what may
be installed upstream of the straight pipe requirement. Often the specifications are given
with little or vague instruction for when installation conditions are not ideal. Figure 1 is
an example of what a manufacturer may describe as an installation requirement:
2
Figure 1. Example Installation Requirement Schematic
Different fixtures have very different effects on the flow profile. In some cases, the
fixtures that apply to the requirement are stated, but not always. This is critical because,
for example, an offset of pipe inside diameter does not carry the same effect as a butterfly
valve at 25% open. The meters are typically tested in the manufacturer’s facilities.
Manufacturers often recommend that additional laboratory calibrations are not necessary.
Disturbances in flow caused by upstream geometry installations can cause errors
in flow measurement. Meters may be installed in a wide variety of circumstances. Pipe
diameter offsets, elbows, valves, and meter orientation are just some of the factors that
affect flow conditions. Kelner (2003) describes installation effects as the potential
deviation in measurement due to such conditions. When space is limited, meters may not
be installed using recommended practices.
Objective
The primary objective of this research was to add quantified data pertaining to the
installation effects of at least one possible pipe setup for magnetic flow meters. For the
meters tested, in many cases the expected accuracy in measurement is not reached, even
3
when meters are installed according to manufacturer specification. The research shows
that calibrations of meters in situ, or in exact replications of the flow conditions they will
be operated in, may be necessary to ensure expected accuracy performance. This is due to
the meter’s sensitivity to changes in the approach velocity profile.
This paper examines the effects of one common installation setup: a meter
oriented vertically (with electrodes at the springline of the pipe) placed downstream of a
single 90° angle elbow oriented horizontally. Laboratory tests were performed with
varying lengths of straight pipe between the meter and elbow. Recommended conditions
were studied based on manufacturer specifications. A range of non-recommended
(shorter) pipe lengths to pipe lengths greater than required were also tested to gain an
overall perspective of the effect of the elbow. The installation effects also are a function
of water velocity. A range of flow velocities up to approximately 16 feet per second were
tested in each pipe setup.
4
CHAPTER II
LITERATURE REVIEW
Several books and studies have provided recommendations on magnetic flow
meter installation practices. The requirements for straight pipe length upstream and
downstream of the meter are either based on empirical testing or on the manufacturer’s
internal studies and recommendations. The figures vary slightly, but fall within the same
general range to achieve a specified level of accuracy. Liptak (2003) gives a conservative
estimate of 3 to 5 diameters upstream and 2 to 3 diameters downstream for accurate
results. Thorn (2000) gives 5 diameters upstream and 5 diameters downstream as a rule
of thumb for straight pipe sections. A study performed by Hanson (1998) suggests 8 to 10
diameters upstream based on information gathered from manufacturers. BS EN 29104
(1993) was referenced for the study conducted by Bates (2000), citing that the meter must
be installed at least 10 diameters from any upstream disturbance and 5 diameters from
any downstream disturbance.
Hanson (1998) performed tests using a range of meter types downstream of a 90°
elbow to show its effect on flow measurement. A pitot tube was used to measure water
velocity profiles across the horizontal pipe diameter. At two diameters, distortion of the
velocity profile was noted with larger velocities corresponding with the outside edge of
the elbow. The flow was shown to normalize for the 5 and 10 diameter tests. Hanson
showed that the distortion had little effect on the error made by the various meters, which
did not include magnetic flow meters. Flow separation (i.e. eddies, vortices) was not
measured in the scope of this study.
5
In the study by Kelner (2003), pipe flow velocity profiles were found to be
affected by installation conditions. Totalized flow measurement was found to be off up to
several percent due to disturbances from an upstream 90° elbow. A long-radius elbow
was used in Kelner’s study, as opposed to a short radius elbow used in this research.
Kelner (2003) states that, in this setup, flow in an ideal state will change to one that has
two counter-rotating vortices and an asymmetric or skewed velocity profile. Disturbances
decay in straight pipe. Pipe wall friction and turbulent diffusion over relatively long
distances without the introduction of additional disturbances causes the decay. In Figure
1, Kelner (2003) illustrates the length of straight pipe required for the effect of the elbow
to no longer persist. Kelner states that, in the conditions of this study, this occurs at
approximately 59 pipe diameters.
Figure 2. Decay of the effect of a single, 90° long radius elbow. (Kelner, 2003)
6
CHAPTER III
EXPERIMENTAL SETUP AND PROCEDURE
Meters
Magnetic flowmeters are sought after for high accuracy and true non-invasive
measurement. Non-invasive refers to the full or near full pipe bore diameter within the
meter itself, allowing for the flow profile to pass undisturbed through the meter with
negligible head loss. Magnetic flowmeters operate based on Faraday’s Law of induction,
which states that if a conductor of length l (meters) is moving at velocity v (meters per
second), perpendicular to a magnetic field of flux density B (Tesla), then a voltage e is
induced across ends of a conductor (Thorn, 2000). This is shown in the equation:
𝑒𝑒 = 𝐵𝐵𝐵𝐵𝐵𝐵
Eq. 1
Applied to the meter, water acts as the conductor moving past sensors within the
meter. The magnetic field is generated by external magnets. Meters selected for this study
were designed with two opposing electrodes located at the springline of the meter to read
the induced voltage. Many of the meters also had additional sensors located at the crown
and invert of the pipe to provide grounding and full pipe status. Alternative electrode
configurations were not a part of this research.
All meters used in this research were donated by anonymous manufacturers. For
the Phase I tests, eleven 10-inch nominal meters were tested. They are distinguished in
the remainder of the paper by using the key in the following table. The meters are listed
along with their manufacturer specifications:
7
Table 1. Meter Accuracy Specifications and Installation Requirements
Requirements
Straight Pipe
Signal
Flow
Meter Key
Type
Accuracy
(fps)
(AC/DC)
Mfr A-1
AC
0.25%
> 1.64
Mfr A-2
DC
0.50%
> 1.64
Mfr B
AC
0.25%
>3
Mfr C-1
AC
0.40%
>3
Mfr C-2
DC
0.40%
>3
Mfr D
AC
0.50%
>0.2
Mfr E
AC
0.50%
>3
Mfr F
AC
0.20%
>1.31
Mfr G
DC
1.00%
>2.84
Mfr H
DC
1.00%
>1.64
Mfr I
DC
0.30%
____ Meters Tested in Phase II
US
(Diam.)
DS
(Diam.)
3
3
5
5
5
1
3
3
2
2
5
2
2
2
3
3
0
2
1
1
2
Single Elbow
US
DS
(Diam.) (Diam.)
3
3
2
2
3
3
2
2
5
2
1
1
For the Phase II tests, a total of four meters were tested. The Mfr B and Mfr F meters
were retested. The Mfr G and Mfr H meters were also tested as new meters of the same
model represented in Phase I testing.
Experimental Setup
Separate procedures, installations, and meters were used for the Phase I and Phase
II test groups. Pipe setups included 10-inch standard wall carbon steel pipe for all testing
except where specified in the setup descriptions. Baseline runs are referred to as the
Straight test in both test groups. All other runs are denoted based on the length of straight
pipe immediately upstream of the meter. For example, “1D” refers to the segment of
straight pipe as one diameter in length (ten inches), “3D” refers to three diameters, and so
8
Figure 3. Phase I Straight Pipe Test Setup (flow left to right)
Figure 4. Phase I 3D Pipe Test Setup (flow left to right)
9
Figure 5. Phase II Straight Test Pipe Setup
forth. Where the meter is installed close coupled with the elbow, the run is denoted as
“CC.”
For Phase I testing, two pipe configurations were installed. The Straight test setup
included 38 feet of straight pipe upstream of the meter location and 7 feet of
straight pipe downstream of the meter location. Additionally, 15 feet of straight 12-inch
pipe was installed upstream of the 10-inch pipe. For the 3D test, a short radius 90°, 10inch horizontal elbow was installed upstream of the meter. The upstream flange of the
meter was installed 30 inches downstream of the elbow flange.
The Phase II Straight test setup included the following pipes and fixtures in order,
starting with the most upstream component:
1. 12-inch vertical long-radius 90° elbow
2.
Butterfly valve
10
3.
12-inch diameter spool which was 39 inches long with a 9 inch long flow
straightener installed on the downstream end
4.
12-inch diameter PVC spool which was 241 inches long
5.
12-inch reference magnetic flowmeter
6.
12-inch diameter spool which was 40 inches long
7.
Reducer which was 18 inches long
8.
10-inch diameter spool which was 201 inches long
The test meter was then installed, with ten diameters of straight pipe on the downstream
side. Visualization of this setup can be found in Figure 5.
The Phase II Elbow test used the same setup up to the 201-inch spool. At that
point a short radius elbow was installed followed by one of five different lengths of pipe
(See Figure 7). The test meter and another section of varying length pipe at least six
diameters long completed the setup.
Figure 6. Phase II Elbow Test Upstream Pipe Setup
11
Figure 7. Variations in Pipe Length for Phase II Elbow Test
Figure 8. Phase II Straight Pipe Test Setup (flow top to bottom)
12
Figure 9. Phase II Elbow Setup for 5D Test (flow left to right)
Procedure
Phase I Test
A laboratory weigh tank was used to measure discharge volume as the
reference for the Phase I test. Regular calibrations are performed on the weigh tank
according to the National Institute of Standards and Technology. The tank capacity is
250,000 pounds. Uncertainty in the results encountered from the tank ranged from
±0.12% to ±0.14%. The flow rate calculated from the volumetric measurement and the
indicated flow rate from the test meter were measured for each run. Flow was calculated
as:
𝑄𝑄 = 𝑊𝑊/(𝑡𝑡 ∗ 𝛾𝛾)
Eq. 2
Where Q is flow rate in cubic feet per second (cfs), W is the weight of water in pounds
13
(lbs), t is time in seconds (s) of the test period, and γ is unit weight of water measured in
pounds per cubic feet (pcf), calculated using a water temperature reading during the test.
A Fluke multimeter was used to average the frequency or 4-20 mA signal, depending on
the meter, from the test meter. For meters recorded using a frequency signal, the flow was
calculated as:
1
𝑄𝑄 = ∗ 𝐻𝐻𝑧𝑧
2
Eq. 3
where Q is flow rate and Hz is the frequency reading (hertz). For milliamp readings, the
flow was calculated as:
𝑄𝑄 =
𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅
16
∗ (𝑚𝑚𝑚𝑚 − 4)
Eq. 4
where Q is flow rate, Range is the range of the flowmeter, and mA is the milliamp
reading. Equations 3 and 4 apply to both the Phase I and Phase II tests. The steps in
obtaining measurement data are outlined as follows:
1. A butterfly valve downstream of the test section was used to set the target flow
rate.
2. The flow was diverted into the weight tank.
3. Simultaneous to the flow diversion, the multimeter was set to average the mag
meter output.
4. Test duration lasted a minimum 180 seconds to reduce sensitivity to random
standard uncertainty.
5. The flow was re-diverted out of the weigh tank.
6. Averaged meter output, weight measurement, and temperature were recorded.
14
Five target flow rates spanning the laboratory range were tested during the
Straight test series. Fifteen target flow rates were tested during the 3D test series. A
second set of data was taken at several flow rates to show repeatability in measurement.
Phase II Test
A 12-inch magnetic flow meter was used to measure discharge volume as the
reference for the Phase II test. The meter was calibrated against the weigh tank used in
Phase I as shown in Figure 10.
During testing, the reference flow rate and the indicated flow rate from the test meter
were measured for each run. Fluke multimeters were used to average the frequency or 420 mA signal, depending on the meter, from both the test meter and the reference meter.
The steps in obtaining measurement data are similar to the Phase I test, substituting the
magnetic meter reading for the weight tank reading. They are outlined as follows:
1. A butterfly valve downstream of the test section was used to set the target flow
rate.
2. The multimeters for each flow meter were set to average the flow rate
simultaneously.
3. Test duration lasted a minimum 180 seconds, based on the flow rate, to ensure
accurate results.
4. Averaged meter outputs were recorded and compared.
15
Figure 10. Reference Meter Calibration
16
CHAPTER IV
RESULTS AND DISCUSSION
The data are represented graphically for each specific meter used in the accuracy
tests. General use of the results may be applied to the category of magnetic flow meters,
but specific results only apply to the meters tested. The range of manufacturers selected
for testing were chosen to show the effect of installation practices for magnetic flow
meters generally. Many of the manufacturers are represented here by only one meter.
Two manufacturers have a second model each represented in the study.
Both the Phase I and Phase II testing are reported as percent deviations from the
reference meter. The reference meter is represented as the horizontal axis at 0.00%.
Manufacturer accuracy and minimum velocity specifications are included as horizontal
and vertical linear bands, respectively. A “passing” status, as used in this report, refers to
the points across all velocities for a given test all falling within the manufacturer’s
accuracy specification.
Above the minimum velocity, it was expected that the data would converge on an
approximate value per test, ideally within the manufacturer’s specification. Maximum
velocities for the range of the meters were not reached in these tests because of laboratory
capacity limitations. All data taken at velocities lower than minimum was done to show
the divergence in accuracy at low velocities.
It would be expected that the further the elbow is away from the meter, the more
accurate the reading will be. This is due to the straight pipe between the two fixtures
17
allowing for velocity profile development. The results verify that, generally, as the meter
is installed closer to the elbow, the deviation from the straight test increases.
Phase I Tests
The test protocol for this series was limited to a baseline (straight pipe) test and a
3D test, despite the varying manufacturer upstream pipe length requirements. The 3D
length was chosen simply to provide information on what performance can be expected
in that condition, rather than match the recommendations for each meter. A common
standard for an upstream length requirement is 3D, though this is not the requirement for
every meter tested. In some cases where the data falls outside the given accuracy
specification, the meter may still perform according to the specification when the
installation requirements are met. On the other hand, some passing results may fall
outside the specification if an upstream pipe length requirement shorter than 3D was
tested.
It can be seen in Figure 11 that Mfr A-1 did not have any of the data fall within
the specifications. The data for the Straight test levels out near the 0.25% specification,
but only near the highest velocities. No data fell inside specifications for the Mfr A-2
meter, as seen in Figure 12. A lot of scatter can be observed in the 3D test, the source of
which is unknown. The only meter where all data fell inside the specifications was Mfr
B, as seen in Figure 13. All data was found to be within ±0.25%. Mfr C-1, seen in Figure
14, saw all data within specification in the Straight test and just outside in the 3D test.
Both data sets show consistent readings across all velocities above minimum. Most of the
Straight test data fell within specification for Mfr C-2, seen in Figure 15. The 3D test
18
shows good consistency with a little scatter across all velocities above minimum. The
Straight test for Mfr E, seen in Figure 17, fell primarily just outside the specification.
Both tests show consistent data across all velocities above minimum. No data fell within
specification for the Mfr F meter, seen in Figure 18. A large amount of scatter and a
negative trend as velocity increased were observed with the Mfr G meter, seen in Figure
19. Only a portion of the data fell within specification. All Straight test data fell within
specification for the Mfr H meter, seen in Figure 20. The 3D showed some data falling
just outside the specification. A negative trend as velocity increases was observed in both
tests. No data fell within specification for the Mfr I meter, seen in Figure 21. Some
scatter was observed at the lower velocities.
Figure 11. Phase I Test Data for Mfr A-1
19
Figure 12. Phase I Test Data for Mfr A-2
Figure 13. Phase I Test Data for Mfr B
20
Figure 14. Phase I Test Data for Mfr C-1
Figure 15. Phase I Test Data for Mfr C-2
21
Figure 16. Phase I Test Data for Mfr D
Figure 17. Phase I Test Data for Mfr E
22
Figure 18. Phase I Test Data for Mfr F
Figure 19. Phase I Test Data for Mfr G
23
Figure 20. Phase I Test Data for Mfr H
Figure 21. Phase I Test Data for Mfr I
24
The Phase I testing can be summarized using the metrics represented in Table 2.
Data recorded at velocities below the manufacturer’s minimum were not included in the
calculated values. The Average Difference from 0% (AD) was calculated as a simple
average of the difference of the observed data point relative to 0% for each meter and
configuration across the velocity range. The Adjusted Average Difference (AAD) was
calculated by shifting the straight test AD value to 0% and shifting the other test by the
same factor. The Max Difference from AD (MD) was calculated as the maximum
difference between each data point and the corresponding AD value. The Standard
Deviation from AD (STDV) was calculated as a typical standard deviation of the AD
value. AD and AAD values are highlighted in red where the average exceeds the
manufacturer specification. AD values highlighted in green refer to all data values falling
within the specification. For the purposes of this report, green values represent “passing”
status for the test.
At first glance, it can be seen that only Mfr B passes both Straight and 3D tests.
The AD values for Mfr G both fall within the manufacturer specification. However, by
looking at the full dataset for Mfr G (Figure 19), it can be seen that many of the data
points actually fell outside of the specification. The upper velocity points bring the
average value within the specification. The MD and STDV can be used as a way of
seeing how the accuracy changes with velocity. The MD and STDV of Mfr G are the
largest of Phase I for both tests, pointing to the least consistency across the range of
velocities.
25
One possible source of scatter in the measurements may be the signal type. The
signal type for each meter can be found in Table 1. Taking a simple average of the MD
and STDV for DC type meters resulted in 0.41% and 0.21% respectively. For AC type
meters, the averages were roughly half of the DC values, at 0.19% and 0.10%
respectively.
Table 2. Phase I Test Statistical Data Summary
Meter
Accy Spec
Test
AD
AAD
MD
STDV
Straight 0.47%
0.00%
0.30%
0.18%
Mfr A±0.25%@3D
1
3D
-2.75% -3.22% 0.29%
0.13%
Straight 3.80%
0.00%
0.18%
0.13%
Mfr A±0.50%@3D
2
3D
1.03% -2.77% 0.56%
0.32%
Straight -0.03% 0.00%
0.13%
0.08%
Mfr B ±0.25%@5D
3D
-0.12% -0.09% 0.17%
0.08%
Straight -0.48% 0.00%
0.09%
0.05%
Mfr C±0.40%@5D
1
3D
0.07%
0.55%
0.10%
0.07%
Straight -0.25% 0.00%
0.26%
0.16%
Mfr C±0.40%@5D
2
3D
0.53%
0.78%
0.22%
0.13%
Straight 0.62%
0.00%
0.17%
0.04%
Mfr D ±0.50%@1D
3D
0.35% -0.26% 0.27%
0.19%
Straight -0.52% 0.00%
0.09%
0.06%
Mfr E ±0.50%@3D
3D
-0.84% -0.32% 0.34%
0.07%
Straight 1.09%
0.00%
0.14%
0.11%
Mfr F ±0.20%@3D
3D
0.83% -0.26% 0.23%
0.12%
Straight 0.76%
0.00%
0.64%
0.39%
Mfr G ±1.00%@2D
3D
0.96%
0.20%
0.96%
0.44%
Straight -0.24% 0.00%
0.45%
0.24%
Mfr H ±1.00%@2D
3D
-1.12% -0.88% 0.34%
0.12%
Straight -1.62% 0.00%
0.23%
0.12%
Mfr I ±0.30%@5D
3D
-0.93% 0.69%
0.26%
0.08%
AD=Average Difference from 0%
AAD=Adjusted Average Difference from 0%
MD=Max Difference from AD
STDV=Standard Deviation of AD
26
Upon visual inspection, the data curves for the 3D and Straight tests for each
meter appear to be separated by a vertical shift in most cases. The Mfr A-2 meter shows
exception due to high levels of scatter in the 3D data. This shows high repeatability
across all testing.
Figure 22 shows how each meter performs at three diameters in relation to a
straight pipe configuration. Each 3D data point is shown as a deviation from the Straight
test, where the Straight data is represented as an average value of the data for that meter.
3D data is represented as a simple difference between each data point and the average
Straight value. The data is corrected, meaning the Straight data is set to 0%. This was
done so that all 3D data could be referenced from the same location. It can be seen that
most meters are within a range of ±1.0% of the straight data.
Figure 22. Phase I 3D Test Deviation From Straight Test
27
Four of the meters read a positive difference, while the other 7 read a negative
difference. In other words, in the 3D elbow test, the meters are more likely to read less
flow rather than more flow, than during the Straight test. The meters retested in Phase II
all read consistently positive or negative at the 3D installation.
Phase II Tests
Plots of Phase II were formatted in a similar fashion to Phase I. It can be
seen that for Mfr F (Figure 23) and Mfr B (Figure 24) neither meter passes “out of the
box.” For this to happen, all points from the installation requirements (3D and 5D,
respectively) would have to fall within the accuracy specification. It can be seen in in the
plot for Mfr G (Figure 25) that all of the data falls within the accuracy specification,
making this the only meter to pass “out of the box.” In Figure 26, the Mfr H meter passes
in the 5D, 10D and Straight tests, but does not pass in the 3D test. The pattern of
deviation suggests that it also may not pass at the installation requirement of 2D. Error
bars were added to each plot to show the maximum range of difference, also given as
values represented above each tested velocity. By looking at the plots for Mfr F (Figure
23) and Mfr B (Figure 24), it can be seen that the maximum difference is consistently
around three percent. This is in contrast with Mfr G (Figure 25) and Mfr H (Figure 26),
where the maximum differences are one to one and a half percent, respectively.
Surprisingly, Mfr H and Mfr G, which have ±1.00% accuracy specifications, had much
narrower bands of deviation across all tests than the two meters with higher standards.
One possibility to this outcome is that the meters with higher standards are
configured to work in more ideal conditions. This is supported by the upstream pipe
28
Figure 23. Phase II Test Data for Mfr F
Figure 24. Phase II Test Data for Mfr B
29
Figure 25. Phase II Test Data for Mfr G
Figure 26. Phase II Test Data for Mfr H
30
length requirement. The lower standard meters both have a requirement of two diameters,
while the higher standard meters have requirements of three and five diameters,
respectively.
The Mfr F and Mfr B meters show, at nearly every velocity, the deviation
increases as upstream pipe length decreases from 10D to close coupled. The Mfr H meter
also showed similarly ordered deviations with the exception of the close-coupled data
sometimes having a smaller deviation than the 1D or even 3D test. The Mfr G meter
generally showed a correlation of shorter upstream pipe length to greater deviation,
though the results showed a more mixed order of pipe lengths in the range of deviation.
In the other three meters, the 10D test showed negligible differences from the Straight
test.
It is unclear from the nature of this study, what exactly is occurring to cause
departure from the length to deviation relationship. Kelner (2003) states that vortices are
formed through flow separation downstream of the elbow. The size and exact location of
the vortices cannot be determined from this study. Changes in the flow profile caused by
the vortex at the location of measurement within the meter could result in a wide range of
uncertainty. In other words, the changes could result in a very large deviation, or none at
all. This could be the case in the example of the Mfr G meter at approximately 4 feet per
second, seen in Figure 25. The 1D and 3D data show a -0.24% and -0.75% deviation,
respectively, from the straight test, ahead of the longer distances of 5D and 10D, at 0.98% and -0.99% respectively. Other sources of noise within the data could include
31
inconsistencies within the meter itself or other unknown distortions in the velocity
profile.
Table 3 summarizes the Phase II data using the same metrics in Table 2 for Phase
I. According to the table, Mfr G passes at all tested lengths. Mfr F passes at the 3D
requirement using AD, though the Mfr B was the only meter to show improvement at the
upstream length requirement, increasing from -0.73% to -0.51%. The meters with
requirements not tested were evaluated based on data from tests with the next length
longer and shorter than the requirement.
Despite the shift from AD to AAD, Mfr G remains the only meter that passes by
its own manufacturer specification. If the remaining meters were to have changes applied
to their installation requirements, the meters could still be claimed to meet the advertised
accuracy. For example, Mfr H does not pass at 2D, as recommended, but would easily
pass if 5D of straight pipe were required instead. Similarly, Mfr B and Mfr F would
require 10D instead of 5D and 3D respectively. The adjusted data was given as an
example, therefore, values were not highlighted.
The standard deviation and max difference are metrics that show how consistent
the meter reads across the range of velocities. From the table, the standard deviations
ranged from 0.11% to 0.37%. Max difference for the set ranged from 0.17% to 0.52%.
Mfr H showed the highest rate of repeatability with an average standard deviation of
0.15%. While Mfr G tied with Mfr F for the highest average standard deviation, it also
carried the highest single standard deviation as well as the highest maximum mean
deviation, making it the least likely to produce repeatable results.
32
Table 3. Phase II Statistical Data Summary
Mfr H
(±1.00%@2D)
Mfr B
(±0.25%@5D)
Mfr G
(±1.00%@2D)
Mfr F
(±0.20%@3D)
Test
AD
AAD
MD
STDV
-0.95% -1.07%
0.19% 0.14%
CC
-1.31% -1.43%
0.20% 0.13%
1
-1.07% -1.19%
0.27% 0.16%
3
-0.28% -0.40%
0.18% 0.17%
5
0.22%
0.10%
0.18% 0.11%
10
0.00%
0.26% 0.15%
Straight 0.12%
0.21% 0.15%
Average
-3.31% -3.09%
0.48% 0.28%
CC
-1.58% -1.36%
0.30% 0.19%
1
-0.89% -0.67%
0.30% 0.18%
3
-0.73% -0.51%
0.24% 0.15%
5
-0.35% -0.13%
0.32% 0.20%
10
0.00%
0.36% 0.21%
Straight -0.22%
0.33% 0.20%
Average
-0.33% -1.01%
0.36% 0.22%
CC
0.13%
-0.55%
0.52% 0.37%
1
0.27%
-0.41%
0.23% 0.15%
3
0.15%
-0.53%
0.24% 0.18%
5
0.63%
-0.05%
0.20% 0.13%
10
0.00%
0.25% 0.17%
Straight 0.68%
0.30% 0.20%
Average
-2.42% -3.12%
0.33% 0.20%
CC
-0.52% -1.22%
0.17% 0.10%
1
0.07%
-0.62%
0.29% 0.17%
3
0.19%
-0.51%
0.16% 0.11%
5
0.76%
0.07%
0.26% 0.17%
10
0.70%
0.00%
0.46% 0.34%
Straight
0.28% 0.18%
Average
AD=Average Difference from 0%
AAD=Adjusted Average Difference from 0%
MD=Max Difference from AD
STDV=Standard Deviation of AD
33
CHAPTER V
CONCLUSION
Meter accuracy is critical to the various applications they are used in. Meters may
not always be installed in ideal circumstances. Information on what can be expected
when non-ideal circumstances occur is highly valuable to the user. This information
provided by the manufacturer does not always match what occurs in the field. Testing
revealed that the distance between an upstream short-radius 90° elbow and a magnetic
flow meter does have an effect on the meter’s accuracy. These results are based on a
sampling of meters made by different manufacturers with varying specifications and
installation requirements.
Results Summary
•
Most of the “off-the-shelf’ meters that were tested during this study did not pass
their given manufacturer specification when installed according to requirements
given by the manufacturer. In fact, most of the meters did not pass the
specification when the meter was installed with 20+ diameters of straight pipe. It
is suggested that a secondary in situ or laboratory calibration be performed to
improve results especially when conditions are non-ideal.
•
When the meters were installed with three diameters between the upstream side of
the meter and a short-radius 90° elbow, it can be expected that the meters will
often read within ±1.0% accuracy of a reading when 20+ diameters of straight
pipe are installed upstream of the meter.
34
•
Decreasing the length of straight pipe upstream of the meter increases the
deviation in flow measurement from the starting reference. Generally, a shift in
accuracy occurs across all flow rates with the change in pipe length. Therefore,
for any velocity, effects on meter accuracy should follow the same expectations
whether the pipe length is at, longer, or shorter than ideal.
•
One factor affecting repeatability may be the type of signal output of the meter. In
Phase I, the DC type meters were found to have roughly twice the average max
difference (MD) and standard deviations (STDV) as the AC type meters.
Need for Further Research
The results for this study would be fortified by testing a more comprehensive
range of meters in a test matrix and pipe setup similar to Phase II. It is recommended that
additional testing is done using similar pipe setups with variations in other parameters.
For example, the meter could be oriented with the sensors vertical rather than horizontal,
or, a double elbow, in or out of plane, could be installed upstream. Research could also be
done on a similar configuration of smaller or larger pipe diameters to see if similar effects
take place. There are innumerable conditions that could be studied, relating to upstream
pipe setup, to better understand their effects on the accuracy of mag meters.
35
REFERENCES
Kelner, E. (2003). “Flow meter installation effects.” <http://asgmt.com/wpcontent/uploads/pdf-docs/2007/1/033.pdf> (Nov. 15, 2015).
Hanson, B. R., and Schwankl, L. J. (1998). “Error Analysis of Flowmeter
Measurements.” J. Irrig. Drain Eng., 124(5), 248–256.
Bates, C. (2000). “Performance of two electromagnetic flowmeters mounted downstream
of a 90° mitre bend/reducer combination.” Measurement, 27(3), 197–206.
Liptak, B. (2003). Instrument Engineers' Handbook Chapter 2. Flow Measurement.
(Fourth ed., Vol. One, pp. 151-400). CRC Press, Florida.
Thorn, R., Melling, A., Kochner, H., Haak, R., Husain, Z., & Wass, D. … (1999). The
Measurement, Instrumentation and Sensors Handbook. Chapter 28. Flow Measurement.
CRC Press, Florida.